1
00:00:00,233 --> 00:00:01,300
悟性是什么

2
00:00:01,333 --> 00:00:02,433
顿悟是干嘛

3
00:00:02,600 --> 00:00:04,000
神经科学为你解答

4
00:00:04,566 --> 00:00:07,166
先给大家讲一个很常见的场景

5
00:00:07,166 --> 00:00:08,166
两个人读同一本书

6
00:00:08,966 --> 00:00:10,100
一个人看完说

7
00:00:10,166 --> 00:00:11,366
挺有道理的

8
00:00:11,400 --> 00:00:13,000
然后把书放下了

9
00:00:13,000 --> 00:00:13,800
该干嘛干嘛

10
00:00:14,533 --> 00:00:17,500
另一个人看到某一页的时候突然愣住了

11
00:00:17,533 --> 00:00:18,666
心跳加速

12
00:00:18,700 --> 00:00:19,733
瞳孔放大

13
00:00:19,733 --> 00:00:23,633
一种电流从脊椎底部蹿上来直达头顶

14
00:00:23,633 --> 00:00:24,533
他放下书

15
00:00:24,566 --> 00:00:25,666
闭上眼

16
00:00:25,700 --> 00:00:27,700
觉得自己好像突然看懂了一些

17
00:00:27,700 --> 00:00:29,233
以前怎么都想不通的事情

18
00:00:30,066 --> 00:00:31,100
从那天起

19
00:00:31,133 --> 00:00:31,500
他变了

20
00:00:32,066 --> 00:00:33,333
这两个人的差距

21
00:00:33,366 --> 00:00:35,633
我们通常用一个词来概括

22
00:00:35,666 --> 00:00:36,233
悟性

23
00:00:36,700 --> 00:00:38,266
但悟性到底是什么

24
00:00:38,500 --> 00:00:40,233
是基因决定的天赋

25
00:00:40,366 --> 00:00:42,600
是智商的另一种说法

26
00:00:42,700 --> 00:00:44,200
还是一种可以训练

27
00:00:44,200 --> 00:00:45,800
可以提高的能力

28
00:00:45,933 --> 00:00:47,700
如果悟性可以提高

29
00:00:47,700 --> 00:00:49,500
它的训练方法到底是什么

30
00:00:50,000 --> 00:00:51,466
在今天这期视频里

31
00:00:51,500 --> 00:00:54,733
我要用2025年最新的神经科学研究

32
00:00:54,733 --> 00:00:55,666
来回答这些问题

33
00:00:56,600 --> 00:01:00,833
但我先剧透一下让我最兴奋的结论：悟性的高低

34
00:01:00,833 --> 00:01:03,933
本质上不取决于你的大脑有多聪明

35
00:01:03,933 --> 00:01:06,533
而取决于你的大脑有多愿意被改变

36
00:01:07,233 --> 00:01:07,866
哈喽

37
00:01:07,900 --> 00:01:08,566
大家好

38
00:01:08,566 --> 00:01:09,166
我是王利杰

39
00:01:09,666 --> 00:01:10,766
要理解悟性

40
00:01:10,800 --> 00:01:13,133
我们得先搞清楚顿悟这件事

41
00:01:13,133 --> 00:01:15,066
在大脑里到底发生了什么

42
00:01:15,600 --> 00:01:17,133
2025年5月

43
00:01:17,133 --> 00:01:18,000
柏林洪堡大学

44
00:01:18,800 --> 00:01:23,366
汉堡大学与杜克大学的联合认知神经科学团队

45
00:01:23,400 --> 00:01:26,033
在<自然·通讯>上发表了一篇非常重要的论文

46
00:01:26,700 --> 00:01:28,766
他们做了一个巧妙的实验

47
00:01:28,800 --> 00:01:31,500
让参与者在核磁共振仪里

48
00:01:31,533 --> 00:01:33,666
看一种叫做Mooney图片的视觉素材

49
00:01:34,466 --> 00:01:35,500
这种图片

50
00:01:35,533 --> 00:01:39,033
就是把一张正常照片的对比度拉到极端

51
00:01:39,033 --> 00:01:40,966
变成纯黑纯白的色块

52
00:01:41,900 --> 00:01:44,700
你第一眼看上去完全是一团混乱

53
00:01:44,700 --> 00:01:46,200
根本不知道这是什么

54
00:01:46,633 --> 00:01:48,000
但在某个瞬间

55
00:01:48,033 --> 00:01:50,733
你的大脑突然一下把碎片拼起来了

56
00:01:50,733 --> 00:01:52,133
你看到了一只狗

57
00:01:52,166 --> 00:01:53,333
或者一个咖啡杯

58
00:01:54,100 --> 00:01:54,966
这个瞬间

59
00:01:54,966 --> 00:01:56,266
就是一次微型的顿悟

60
00:01:56,933 --> 00:01:58,866
实验结果有三个关键发现

61
00:01:59,533 --> 00:02:00,533
第一

62
00:02:00,566 --> 00:02:02,066
顿悟的那一瞬间

63
00:02:02,066 --> 00:02:04,133
参与者大脑的视觉皮层

64
00:02:04,166 --> 00:02:06,600
发生了剧烈的表征重组

65
00:02:06,633 --> 00:02:10,366
他们的神经元不是在原有的模式上做了微调

66
00:02:10,366 --> 00:02:13,066
而是整个激活模式发生了结构性的切换

67
00:02:13,833 --> 00:02:15,066
这不是渐变

68
00:02:15,100 --> 00:02:15,800
是突变

69
00:02:16,666 --> 00:02:18,166
用一个类比来说

70
00:02:18,200 --> 00:02:21,500
这不是把收音机的音量调大了一格

71
00:02:21,500 --> 00:02:22,933
而是整个换了一个频道

72
00:02:23,600 --> 00:02:24,400
第二

73
00:02:24,400 --> 00:02:26,300
在表征重组的同时

74
00:02:26,300 --> 00:02:29,466
大脑的海马体和杏仁核被强烈激活了

75
00:02:30,233 --> 00:02:31,833
海马体负责记忆

76
00:02:31,833 --> 00:02:33,733
杏仁核负责情绪

77
00:02:34,066 --> 00:02:36,966
它们和视觉皮层之间的同步性急剧上升

78
00:02:37,566 --> 00:02:38,600
也就是说

79
00:02:38,600 --> 00:02:39,833
顿悟的时候

80
00:02:39,866 --> 00:02:40,700
认知

81
00:02:40,700 --> 00:02:43,966
情绪和记忆三个系统同时开火

82
00:02:44,000 --> 00:02:45,700
形成了一个统一的神经网络

83
00:02:46,966 --> 00:02:47,666
第三

84
00:02:47,666 --> 00:02:49,200
也是最让我震撼的

85
00:02:49,533 --> 00:02:51,433
通过顿悟获得的答案

86
00:02:51,466 --> 00:02:54,633
记忆持久度是渐进式学习的将近两倍

87
00:02:55,633 --> 00:02:57,566
Cabeza教授的原话是

88
00:02:57,566 --> 00:03:01,200
如果你在学习某样东西的时候经历了一次顿悟

89
00:03:01,233 --> 00:03:03,366
你的记忆力几乎翻倍

90
00:03:03,366 --> 00:03:06,200
几乎没有什么记忆效应能比这个更强大了

91
00:03:06,866 --> 00:03:08,900
你知道这意味着什么吗

92
00:03:09,000 --> 00:03:11,933
这意味着顿悟不只是感觉上很爽

93
00:03:11,966 --> 00:03:16,266
它在物理层面上重塑了你大脑的信息表征方式

94
00:03:16,300 --> 00:03:18,200
并且把这个新的表征

95
00:03:18,200 --> 00:03:19,733
深深地烙进了长期记忆里

96
00:03:20,400 --> 00:03:20,933
好

97
00:03:21,266 --> 00:03:22,766
记住这三个发现

98
00:03:22,800 --> 00:03:25,300
因为接下来我要把这个实验结果

99
00:03:25,300 --> 00:03:28,400
翻译成我们频道一直在用的预测编码框架

100
00:03:29,333 --> 00:03:32,866
还记得我在打坐到底在干什么那期讲过的内容吗

101
00:03:32,966 --> 00:03:35,700
你的大脑不是被动的信息接收器

102
00:03:35,700 --> 00:03:37,633
而是一台主动的预测发生器

103
00:03:38,500 --> 00:03:41,033
就像现在的大语言模型一样

104
00:03:41,033 --> 00:03:45,300
它时时刻刻都在自上而下地生成对世界的预测

105
00:03:45,333 --> 00:03:48,866
感官数据只传回预测和现实之间的差值

106
00:03:49,033 --> 00:03:50,066
也就是预测误差

107
00:03:50,933 --> 00:03:54,233
而每一个先验信念都有一个精度权重

108
00:03:54,233 --> 00:03:56,833
你的大脑对这个信念有多确信

109
00:03:56,866 --> 00:03:58,100
这个权重就有多高

110
00:03:58,833 --> 00:04:00,566
现在让我问你一个问题

111
00:04:00,566 --> 00:04:02,366
在预测编码的框架里

112
00:04:02,366 --> 00:04:04,000
顿悟到底是什么

113
00:04:04,100 --> 00:04:05,700
答案非常清晰

114
00:04:05,700 --> 00:04:06,300
顿悟

115
00:04:06,333 --> 00:04:09,700
就是一次大规模的先验信念更新事件

116
00:04:10,233 --> 00:04:12,166
你看那张Mooney图片的时候

117
00:04:12,166 --> 00:04:15,966
你的大脑一开始在用旧的先验信念去解读它

118
00:04:16,266 --> 00:04:18,433
这是一堆随机的黑白色块

119
00:04:18,966 --> 00:04:21,066
这个先验信念生成的预测

120
00:04:21,066 --> 00:04:24,733
和实际视觉输入之间存在巨大的预测误差

121
00:04:24,766 --> 00:04:28,833
但你的大脑暂时找不到更好的模型来消除这个误差

122
00:04:29,333 --> 00:04:31,800
于是误差信号一直在累积

123
00:04:31,800 --> 00:04:33,300
在你意识不到的地方

124
00:04:33,300 --> 00:04:35,233
不断地冲击着旧模型的墙壁

125
00:04:36,266 --> 00:04:38,000
然后在某个临界点

126
00:04:38,033 --> 00:04:38,333
墙塌了

127
00:04:39,233 --> 00:04:41,166
一个全新的先验信念

128
00:04:41,166 --> 00:04:42,300
这是一只狗

129
00:04:42,333 --> 00:04:42,866
突然涌现

130
00:04:43,700 --> 00:04:47,133
这个新模型一下子消除了绝大部分预测误差

131
00:04:47,933 --> 00:04:51,266
误差的瀑布式下降触发了多巴胺释放

132
00:04:51,300 --> 00:04:52,933
这就是那股快感的来源

133
00:04:53,800 --> 00:04:55,800
海马体同时被激活

134
00:04:55,800 --> 00:04:58,433
把这个新模型的整个涌现过程

135
00:04:58,466 --> 00:05:01,300
和对应的情绪状态一起打包

136
00:05:01,300 --> 00:05:02,233
存入长期记忆

137
00:05:03,066 --> 00:05:06,033
所以Becker实验里看到的表征重组

138
00:05:06,066 --> 00:05:08,366
用预测编码的语言翻译就是

139
00:05:08,400 --> 00:05:12,233
大脑的内部模型发生了一次不可逆的结构性跃迁

140
00:05:13,733 --> 00:05:15,233
旧的先验信念

141
00:05:15,266 --> 00:05:16,666
被一个全新的

142
00:05:16,666 --> 00:05:18,366
更高效的先验信念替代了

143
00:05:19,133 --> 00:05:19,866
你发现了吗

144
00:05:20,366 --> 00:05:21,833
这和物理学里的相变

145
00:05:22,300 --> 00:05:23,800
逻辑上是完全一致的

146
00:05:24,733 --> 00:05:26,533
水在零度以下是冰

147
00:05:26,566 --> 00:05:28,200
你给它加热

148
00:05:28,200 --> 00:05:30,266
温度一度一度地升高

149
00:05:30,300 --> 00:05:34,566
冰的内部结构在微观层面缓慢地积累着变化

150
00:05:34,566 --> 00:05:35,766
但从外面看

151
00:05:35,800 --> 00:05:36,733
它还是冰

152
00:05:37,200 --> 00:05:39,266
直到到了零度的那一瞬间

153
00:05:39,266 --> 00:05:40,300
相变发生了

154
00:05:40,966 --> 00:05:43,000
冰不是慢慢变成水的

155
00:05:43,033 --> 00:05:44,933
它是在一个临界点上

156
00:05:44,933 --> 00:05:46,966
突然跳到了一个全新的状态

157
00:05:47,433 --> 00:05:48,166
顿悟

158
00:05:48,166 --> 00:05:49,700
就是大脑的相变

159
00:05:50,266 --> 00:05:53,433
而且我在<万亿参数>那期讲过的，AI

160
00:05:53,466 --> 00:05:55,200
领域的Grokking现象

161
00:05:55,200 --> 00:05:58,600
和这个机制在数学上几乎完全一致

162
00:05:58,966 --> 00:06:02,433
神经网络在训练初期只是在死记硬背

163
00:06:02,433 --> 00:06:04,466
误差曲线下降得很慢

164
00:06:04,833 --> 00:06:08,400
但在数据量和训练时间积累到某个临界点之后

165
00:06:08,666 --> 00:06:09,766
突然之间

166
00:06:09,766 --> 00:06:12,066
误差悬崖式下跌

167
00:06:12,066 --> 00:06:14,133
模型不再是机械记忆

168
00:06:14,166 --> 00:06:17,133
而是突然懂了背后的通用规则

169
00:06:17,466 --> 00:06:18,366
那个悟

170
00:06:18,366 --> 00:06:19,333
在数学上

171
00:06:19,366 --> 00:06:21,800
就是参数在高维空间里

172
00:06:21,833 --> 00:06:25,633
从混乱状态坍缩成了一个低熵的几何结构

173
00:06:25,633 --> 00:06:29,000
这和Becker实验里看到的大脑表征重组

174
00:06:29,000 --> 00:06:31,800
是同一个现象的不同尺度表达

175
00:06:32,033 --> 00:06:32,666
好

176
00:06:32,700 --> 00:06:34,833
如果顿悟是大脑的相变

177
00:06:34,866 --> 00:06:37,333
那悟性高意味着什么

178
00:06:37,433 --> 00:06:38,366
悟性高

179
00:06:38,366 --> 00:06:40,966
不是说你相变的结果比别人好

180
00:06:40,966 --> 00:06:44,566
不是说你看到的那只狗比别人的更清晰

181
00:06:45,133 --> 00:06:46,100
悟性高

182
00:06:46,100 --> 00:06:49,500
是说你的大脑到达临界点的阈值比别人低

183
00:06:49,900 --> 00:06:51,300
同样的信息输入

184
00:06:51,333 --> 00:06:53,933
你的大脑更容易发生相变

185
00:06:54,400 --> 00:06:56,033
那问题就变成了

186
00:06:56,066 --> 00:06:58,500
是什么决定了这个阈值的高低

187
00:06:58,733 --> 00:07:01,366
答案就藏在预测编码理论里

188
00:07:01,400 --> 00:07:02,600
精度权重

189
00:07:02,866 --> 00:07:05,833
你的旧先验信念的精度权重越高

190
00:07:05,866 --> 00:07:08,400
你的大脑就越不愿意更新它

191
00:07:09,566 --> 00:07:11,200
因为精度权重高

192
00:07:11,200 --> 00:07:14,833
意味着大脑认为这个信念非常可靠

193
00:07:14,833 --> 00:07:17,466
即使收到了大量的预测误差信号

194
00:07:17,500 --> 00:07:19,933
大脑也会倾向于解释掉它们

195
00:07:19,933 --> 00:07:21,666
而不是更新信念

196
00:07:21,966 --> 00:07:24,766
这就是为什么固执的人很少有顿悟

197
00:07:24,766 --> 00:07:26,800
他们的先验权重太高了

198
00:07:26,800 --> 00:07:29,533
高到预测误差信号根本穿不透

199
00:07:29,933 --> 00:07:30,800
反过来

200
00:07:30,833 --> 00:07:32,166
一个悟性高的人

201
00:07:32,166 --> 00:07:34,466
他的先验信念的精度权重

202
00:07:34,466 --> 00:07:36,633
处于一个微妙的平衡状态

203
00:07:36,633 --> 00:07:39,600
高到足以让他形成连贯的世界模型

204
00:07:39,633 --> 00:07:42,533
但又低到足以让新的预测误差信号

205
00:07:42,533 --> 00:07:45,300
有机会冲击和更新这个模型

206
00:07:45,600 --> 00:07:47,733
想象你用沙子堆了一座城堡

207
00:07:48,166 --> 00:07:50,933
如果是钢筋混凝土浇筑的

208
00:07:50,933 --> 00:07:52,833
海浪拍过来毫发无损

209
00:07:52,833 --> 00:07:53,933
那就是固执

210
00:07:54,233 --> 00:07:55,766
如果是散沙一盘

211
00:07:55,800 --> 00:07:57,033
风一吹就没了

212
00:07:57,066 --> 00:07:58,166
那就是混乱

213
00:07:58,466 --> 00:07:59,666
悟性是什么

214
00:07:59,833 --> 00:08:01,400
是沙堡刚好站着

215
00:08:01,400 --> 00:08:02,966
但海浪来的时候

216
00:08:02,966 --> 00:08:05,200
它愿意被重塑成新的形状

217
00:08:05,633 --> 00:08:06,633
你有没有发现

218
00:08:06,666 --> 00:08:09,900
这和我们在<涌现>那期讲过的自组织临界性

219
00:08:09,900 --> 00:08:11,166
是完全一致的

220
00:08:11,833 --> 00:08:14,400
巴克和他的沙堆模型告诉我们

221
00:08:14,400 --> 00:08:16,366
复杂系统在临界态附近

222
00:08:16,366 --> 00:08:18,866
最容易产生大规模的级联反应

223
00:08:19,366 --> 00:08:20,566
沙堆太平坦

224
00:08:20,566 --> 00:08:22,866
加一粒沙什么都不会发生

225
00:08:23,466 --> 00:08:24,633
沙堆太陡峭

226
00:08:24,666 --> 00:08:25,966
它已经崩过了

227
00:08:25,966 --> 00:08:27,166
处于稳定态

228
00:08:27,466 --> 00:08:30,266
只有在刚好处于临界点的状态

229
00:08:30,266 --> 00:08:32,466
一粒沙就能引发雪崩

230
00:08:33,166 --> 00:08:33,733
悟性

231
00:08:33,766 --> 00:08:36,966
就是你的认知系统维持在临界态附近的能力

232
00:08:37,200 --> 00:08:39,133
这就解释了为什么同一本书

233
00:08:39,166 --> 00:08:40,500
同一句话

234
00:08:40,500 --> 00:08:43,133
对不同的人效果天差地别

235
00:08:43,666 --> 00:08:46,733
六祖慧能听到应无所住而生其心

236
00:08:46,733 --> 00:08:48,433
这八个字就开悟了

237
00:08:49,233 --> 00:08:50,766
这八个字你也读过

238
00:08:50,766 --> 00:08:52,033
我也读过

239
00:08:52,066 --> 00:08:53,700
为什么我们没开悟

240
00:08:53,833 --> 00:08:55,566
不是因为我们的智商不够

241
00:08:55,566 --> 00:08:59,100
而是因为我们的认知沙堆还没有堆到临界点

242
00:08:59,866 --> 00:09:01,866
慧能在听到这句话之前

243
00:09:01,866 --> 00:09:03,200
他的人生经历

244
00:09:03,233 --> 00:09:04,600
他的内在探索

245
00:09:04,600 --> 00:09:06,200
他对痛苦的直面

246
00:09:06,200 --> 00:09:09,266
已经把他的认知系统推到了悬崖边缘

247
00:09:09,933 --> 00:09:11,066
八个字落下来

248
00:09:11,066 --> 00:09:12,233
就是最后那粒沙

249
00:09:13,900 --> 00:09:14,700
当然

250
00:09:14,700 --> 00:09:16,266
佛学意义上的开悟

251
00:09:16,300 --> 00:09:18,333
远不止一次认知上的顿悟

252
00:09:18,566 --> 00:09:20,833
但那个临界瞬间的核心逻辑

253
00:09:20,866 --> 00:09:21,800
是相通的

254
00:09:22,266 --> 00:09:23,166
所以你看

255
00:09:23,366 --> 00:09:25,166
悟性不是凭空的天赋

256
00:09:25,466 --> 00:09:27,466
它是你的整个人生经历

257
00:09:27,466 --> 00:09:31,533
在认知参数空间里积累出来的一种特定的拓扑状态

258
00:09:32,166 --> 00:09:35,166
先天的基因会影响我们的认知起点

259
00:09:35,533 --> 00:09:37,900
它决定了初始的认知地形

260
00:09:37,900 --> 00:09:41,000
但它锁死不了你能不能走到相变的临界点

261
00:09:41,766 --> 00:09:43,233
路的长短不同

262
00:09:43,233 --> 00:09:44,233
但路没有断过

263
00:09:45,300 --> 00:09:48,366
这也回答了我们频道之前提出的那个核心共识

264
00:09:48,866 --> 00:09:51,800
文本解读的天花板不是语言能力

265
00:09:51,833 --> 00:09:52,700
而是体验维度

266
00:09:53,566 --> 00:09:54,733
同一段文字

267
00:09:54,766 --> 00:09:57,100
一个经历过至亲离世的人

268
00:09:57,100 --> 00:09:59,200
和一个从未面对过死亡的人

269
00:09:59,233 --> 00:10:00,966
读出来的东西是完全不同的

270
00:10:01,866 --> 00:10:04,000
不是因为后者不认字

271
00:10:04,000 --> 00:10:06,600
而是因为他的认知参数空间里

272
00:10:06,600 --> 00:10:07,600
缺少某些维度

273
00:10:08,266 --> 00:10:11,333
那些维度只有通过真实的体验才能被打开

274
00:10:12,266 --> 00:10:13,400
一旦打开

275
00:10:13,433 --> 00:10:17,933
同一段文字就能在更高维度的空间里引发共振

276
00:10:17,933 --> 00:10:20,466
而共振达到临界强度的时候

277
00:10:20,466 --> 00:10:21,300
相变就发生了

278
00:10:22,233 --> 00:10:22,866
好

279
00:10:22,866 --> 00:10:27,133
如果悟性的本质是认知系统在临界态附近的灵活性

280
00:10:27,166 --> 00:10:28,333
那怎么提高它

281
00:10:28,866 --> 00:10:29,800
我总结了四个层面

282
00:10:30,433 --> 00:10:31,633
第一个层面

283
00:10:31,633 --> 00:10:33,033
堆积沙堆

284
00:10:33,066 --> 00:10:34,333
扩展你的体验维度

285
00:10:35,200 --> 00:10:35,900
这个很直观

286
00:10:36,533 --> 00:10:38,433
你的认知沙堆越高

287
00:10:38,500 --> 00:10:39,366
越多维

288
00:10:39,433 --> 00:10:40,633
你就越接近临界点

289
00:10:41,600 --> 00:10:42,400
读书

290
00:10:42,433 --> 00:10:43,333
旅行

291
00:10:43,333 --> 00:10:46,333
和不同领域的人深度对话

292
00:10:46,333 --> 00:10:48,166
学习一门全新的技能

293
00:10:48,166 --> 00:10:50,666
甚至经历一次重大的挫折

294
00:10:50,666 --> 00:10:53,033
所有这些都在往你的沙堆上加沙子

295
00:10:54,133 --> 00:10:54,966
但注意

296
00:10:55,000 --> 00:10:56,500
这里有一个陷阱

297
00:10:56,500 --> 00:10:59,133
如果你只在同一个领域里堆沙子

298
00:10:59,133 --> 00:11:02,600
你的沙堆可能会变得很高但很窄

299
00:11:02,600 --> 00:11:06,133
它在那个方向上早就过了临界点崩过了

300
00:11:06,166 --> 00:11:07,400
处于稳定态

301
00:11:07,433 --> 00:11:09,366
再加沙子也不会引发雪崩

302
00:11:10,433 --> 00:11:14,166
真正有效的做法是跨维度堆沙子

303
00:11:14,166 --> 00:11:15,333
你懂物理学

304
00:11:15,366 --> 00:11:16,666
你也读佛学

305
00:11:16,666 --> 00:11:18,500
你还做过投资

306
00:11:18,533 --> 00:11:21,200
研究过大语言模型的技术原理

307
00:11:21,233 --> 00:11:23,333
你还亲手写过代码

308
00:11:23,333 --> 00:11:26,700
这些来自完全不同领域的沙子堆在一起

309
00:11:26,733 --> 00:11:28,400
更容易把系统推到临界态

310
00:11:29,500 --> 00:11:32,133
这也是为什么历史上最伟大的顿悟

311
00:11:32,133 --> 00:11:34,000
往往发生在学科交叉的地方

312
00:11:34,600 --> 00:11:35,366
第二个层面

313
00:11:35,633 --> 00:11:36,900
调低精度权重

314
00:11:37,333 --> 00:11:38,466
练习认知松绑

315
00:11:39,133 --> 00:11:42,800
这就是我们在打坐那期反复讲的修行的核心机制

316
00:11:43,600 --> 00:11:44,933
你可以感受一下

317
00:11:44,966 --> 00:11:47,766
你此刻最坚定的一个信念是什么

318
00:11:47,866 --> 00:11:49,066
不管它是什么

319
00:11:49,066 --> 00:11:50,233
你注意到了吗

320
00:11:50,233 --> 00:11:52,333
当我让你去审视它的时候

321
00:11:52,333 --> 00:11:54,300
你的身体有一个微妙的紧缩

322
00:11:54,666 --> 00:11:56,700
好像有什么东西被威胁到了

323
00:11:57,433 --> 00:11:58,300
那个紧缩

324
00:11:58,600 --> 00:12:00,933
就是精度权重在身体上的表达

325
00:12:01,966 --> 00:12:05,433
你的很多先验信念的精度权重被调得太高了

326
00:12:05,466 --> 00:12:06,700
我必须这样理解

327
00:12:06,966 --> 00:12:08,366
世界一定是那样的

328
00:12:08,533 --> 00:12:09,633
这个理论不可能错

329
00:12:11,000 --> 00:12:15,000
这些过高的权重像钢筋一样加固了你的旧模型

330
00:12:15,033 --> 00:12:17,366
让新的预测误差信号打不进来

331
00:12:17,600 --> 00:12:18,800
怎么调低

332
00:12:18,900 --> 00:12:22,100
冥想是一条被大量研究验证的路径

333
00:12:22,466 --> 00:12:25,266
大量针对冥想的神经影像学研究

334
00:12:25,266 --> 00:12:29,400
和基于预测编码框架的相关发现都指向同一个结论

335
00:12:29,833 --> 00:12:31,166
深度冥想时

336
00:12:31,166 --> 00:12:34,800
大脑高层先验信念的精度权重会被降低

337
00:12:35,300 --> 00:12:37,733
默认模式网络的活动下降

338
00:12:37,766 --> 00:12:40,400
整个大脑会进入一种更开放

339
00:12:40,400 --> 00:12:41,600
更灵活的状态

340
00:12:42,100 --> 00:12:43,700
但冥想不是唯一的方法

341
00:12:44,366 --> 00:12:48,133
任何让你暂时搁置我知道了这个感觉的练习

342
00:12:48,166 --> 00:12:49,866
都在做同样的事情

343
00:12:50,533 --> 00:12:51,133
比如

344
00:12:51,133 --> 00:12:54,633
当你读到一个和你观点完全相反的论述时

345
00:12:54,633 --> 00:12:55,866
试着不急于反驳

346
00:12:55,866 --> 00:12:57,600
而是先问自己

347
00:12:57,633 --> 00:12:58,766
如果他是对的

348
00:12:58,766 --> 00:13:01,200
我的模型哪里需要修改

349
00:13:01,333 --> 00:13:02,800
这个简单的练习

350
00:13:02,800 --> 00:13:05,866
就是在手动调低你的先验权重

351
00:13:06,166 --> 00:13:07,466
第三个层面

352
00:13:07,500 --> 00:13:09,166
增加预测误差的信号强度

353
00:13:09,966 --> 00:13:10,933
前面讲了

354
00:13:10,966 --> 00:13:14,666
宇航员在太空看地球之所以能引发总观效应

355
00:13:15,266 --> 00:13:18,666
是因为那个视觉刺激的精度权重被拉到了最高

356
00:13:19,166 --> 00:13:20,566
全感官沉浸

357
00:13:20,566 --> 00:13:22,300
生存性危机加持

358
00:13:22,300 --> 00:13:24,433
没有任何注意力竞争性刺激

359
00:13:24,800 --> 00:13:27,100
同样的信息在电视上看

360
00:13:27,133 --> 00:13:28,500
精度权重极低

361
00:13:28,766 --> 00:13:30,666
根本打不动先验信念

362
00:13:31,066 --> 00:13:32,800
这对日常学习的启示是

363
00:13:33,133 --> 00:13:34,700
你获取信息的方式

364
00:13:35,033 --> 00:13:36,866
和信息本身同样重要

365
00:13:37,433 --> 00:13:39,166
同一本书摆在那里

366
00:13:39,166 --> 00:13:41,200
你是在沙发上随便翻翻

367
00:13:41,400 --> 00:13:44,533
还是因为你正在经历一个真实的困境

368
00:13:44,566 --> 00:13:47,566
带着一个烙在心里最深处的问题去读

369
00:13:47,966 --> 00:13:49,200
同样的信息

370
00:13:49,233 --> 00:13:51,100
当你带着真实的困惑

371
00:13:51,100 --> 00:13:52,433
真实的需要

372
00:13:52,433 --> 00:13:55,133
真实的痛苦去接触它的时候

373
00:13:55,166 --> 00:13:58,600
你的大脑给它赋予的精度权重完全不同

374
00:13:59,266 --> 00:14:01,666
这就是为什么带着问题学习

375
00:14:01,700 --> 00:14:03,733
比泛泛而学有效得多

376
00:14:03,766 --> 00:14:05,533
不是因为你更努力了

377
00:14:05,566 --> 00:14:08,733
而是因为你的大脑从神经机制层面

378
00:14:08,733 --> 00:14:11,200
就更认真地对待了这些输入

379
00:14:11,300 --> 00:14:12,433
第四个层面

380
00:14:12,466 --> 00:14:14,000
也是最深的一层

381
00:14:14,033 --> 00:14:15,133
让元神介入

382
00:14:15,700 --> 00:14:18,000
我在讲<脑海里有两个你

383
00:14:18,000 --> 00:14:19,000
一个在说

384
00:14:19,000 --> 00:14:21,666
一个在听>那期视频里提过

385
00:14:21,666 --> 00:14:24,533
道家讲的识神就是那个不停说话

386
00:14:24,566 --> 00:14:25,700
不停分析

387
00:14:25,733 --> 00:14:28,266
不停判断的本地智能小模型

388
00:14:28,766 --> 00:14:31,400
元神是当所有噪音平息之后

389
00:14:31,400 --> 00:14:34,466
才浮现出来的更深层的觉知系统

390
00:14:34,500 --> 00:14:35,700
你的觉性

391
00:14:35,733 --> 00:14:37,833
或则说你的如来藏

392
00:14:38,200 --> 00:14:41,166
很多伟大的顿悟不是靠想出来的

393
00:14:41,166 --> 00:14:42,300
恰恰相反

394
00:14:42,333 --> 00:14:44,966
它们是在你停止思考的时候出现的

395
00:14:45,900 --> 00:14:48,233
阿基米德是在泡澡的时候

396
00:14:48,266 --> 00:14:51,900
想到了用排水法测算不规则物体的体积

397
00:14:51,900 --> 00:14:54,533
由此解开了王冠掺假的难题

398
00:14:54,566 --> 00:14:57,833
也为后来的浮力原理奠定了核心基础

399
00:14:58,300 --> 00:15:00,766
凯库勒是在打瞌睡的时候

400
00:15:00,766 --> 00:15:02,166
梦到了苯环结构

401
00:15:02,766 --> 00:15:06,266
庞加莱是在踏上公共汽车台阶的那一瞬间

402
00:15:06,300 --> 00:15:07,966
解开了数学难题

403
00:15:08,566 --> 00:15:12,166
拉马努金的几千个数学方程式就更神奇了

404
00:15:12,166 --> 00:15:14,600
我们之前有一期视频专门讲过

405
00:15:15,066 --> 00:15:18,000
这些故事都指向同一个模式

406
00:15:18,033 --> 00:15:19,833
密集的思考阶段之后

407
00:15:19,833 --> 00:15:22,133
一段放松或走神的阶段

408
00:15:22,133 --> 00:15:24,200
然后顿悟突然涌现

409
00:15:24,766 --> 00:15:26,433
用我们的框架翻译

410
00:15:26,466 --> 00:15:29,100
识神在前一个阶段拼命堆沙子

411
00:15:29,133 --> 00:15:30,766
积累预测误差

412
00:15:30,800 --> 00:15:32,566
但精度权重太高

413
00:15:32,566 --> 00:15:34,166
旧模型一直没崩

414
00:15:34,633 --> 00:15:36,066
然后识神累了

415
00:15:36,066 --> 00:15:37,333
暂时停机了

416
00:15:37,866 --> 00:15:41,600
这个时候默认模式网络的独裁控制松动了

417
00:15:41,633 --> 00:15:43,533
元神有机会介入了

418
00:15:43,966 --> 00:15:46,400
元神不像识神那样线性分析

419
00:15:46,400 --> 00:15:47,733
它是全局性的

420
00:15:47,766 --> 00:15:49,033
直觉性的

421
00:15:49,133 --> 00:15:52,300
可以在高维空间里同时看到所有连接

422
00:15:52,766 --> 00:15:54,600
它在那个短暂的窗口里

423
00:15:54,633 --> 00:15:56,966
完成了识神做不到的事情

424
00:15:56,966 --> 00:15:59,133
一次跨维度的模式识别

425
00:15:59,166 --> 00:16:01,266
一次整体性的表征重组

426
00:16:01,866 --> 00:16:05,866
这就是顿悟往往在放松时出现的神经科学解释

427
00:16:05,900 --> 00:16:08,700
不是因为放松本身产生了答案

428
00:16:08,766 --> 00:16:12,300
而是因为放松让识神让出了带宽

429
00:16:12,300 --> 00:16:15,733
让元神终于有机会与高纬的智慧连接

430
00:16:15,766 --> 00:16:16,833
获取信息了

431
00:16:17,366 --> 00:16:19,200
所以如果你想提高悟性

432
00:16:19,233 --> 00:16:21,966
最后一条建议可能是最反直觉的

433
00:16:21,966 --> 00:16:22,833
别一直想

434
00:16:23,266 --> 00:16:25,600
给你的大脑留出不思考的空间

435
00:16:26,033 --> 00:16:26,766
散步

436
00:16:26,800 --> 00:16:27,500
发呆

437
00:16:27,533 --> 00:16:28,233
洗澡

438
00:16:28,233 --> 00:16:29,200
听音乐

439
00:16:29,233 --> 00:16:31,466
做一些不需要动脑的体力活

440
00:16:31,466 --> 00:16:33,700
这些看似浪费时间的事情

441
00:16:33,733 --> 00:16:36,566
恰恰是在给元神创造介入的窗口

442
00:16:37,100 --> 00:16:39,933
最后我想说一个这期内容给我个人最大的触动

443
00:16:40,800 --> 00:16:42,733
我一直觉得悟性这个词很玄

444
00:16:43,400 --> 00:16:45,366
有些人就是一点就通

445
00:16:45,366 --> 00:16:47,200
有些人就是怎么讲都讲不明白

446
00:16:48,100 --> 00:16:51,366
以前我倾向于把这归结为天赋差异

447
00:16:51,400 --> 00:16:52,500
然后就不想了

448
00:16:52,900 --> 00:16:57,166
但当我用预测编码和相变理论重新理解悟性之后

449
00:16:57,166 --> 00:16:59,666
我发现这根本不是天赋问题

450
00:17:00,033 --> 00:17:01,366
它是状态问题

451
00:17:01,866 --> 00:17:03,500
你不是有没有悟性

452
00:17:03,533 --> 00:17:05,466
你是此刻的认知系统

453
00:17:05,466 --> 00:17:07,300
是否处在临界态附近

454
00:17:07,766 --> 00:17:08,800
而这个状态

455
00:17:08,800 --> 00:17:10,200
是可以主动调整的

456
00:17:11,166 --> 00:17:14,200
你可以通过丰富体验维度来堆沙子

457
00:17:14,200 --> 00:17:18,566
你可以通过练习认知松绑来调低僵化的先验权重

458
00:17:18,966 --> 00:17:22,033
你可以通过带着真实问题学习来提高

459
00:17:22,066 --> 00:17:23,866
输入信号的精度权重

460
00:17:24,066 --> 00:17:27,300
你可以通过给大脑留白来让元神介入

461
00:17:27,933 --> 00:17:31,066
悟性不是你出生时被分配的固定属性

462
00:17:31,100 --> 00:17:33,500
跟你的智商和学历都没关系

463
00:17:34,166 --> 00:17:35,133
相反

464
00:17:35,133 --> 00:17:37,033
智商和学历太高

465
00:17:37,033 --> 00:17:39,666
反而容易产生佛学讲的所知障

466
00:17:40,200 --> 00:17:43,433
悟性是你的认知系统在此刻的动态状态

467
00:17:43,433 --> 00:17:44,466
而状态

468
00:17:44,466 --> 00:17:45,366
永远可以改变

469
00:17:45,966 --> 00:17:47,666
这也是为什么佛陀说

470
00:17:47,666 --> 00:17:49,133
众生皆有佛性

471
00:17:49,933 --> 00:17:52,866
佛性不是说你生下来就已经开悟了

472
00:17:53,266 --> 00:17:55,100
而是说你的认知系统

473
00:17:55,133 --> 00:17:57,800
天然地具备发生相变的潜力

474
00:17:58,233 --> 00:18:00,233
临界点永远存在

475
00:18:00,233 --> 00:18:02,433
沙堆永远可以继续堆

476
00:18:02,433 --> 00:18:04,800
精度权重永远可以调

477
00:18:05,333 --> 00:18:06,566
问题只在于

478
00:18:06,566 --> 00:18:09,566
你愿不愿意让自己的旧模型被打碎

479
00:18:09,600 --> 00:18:13,333
愿不愿意承受那个相变前夜的混乱与不安

480
00:18:13,633 --> 00:18:15,000
最深刻的顿悟

481
00:18:15,033 --> 00:18:17,366
往往发生在最痛苦的时刻

482
00:18:17,833 --> 00:18:20,366
不是因为痛苦本身有什么神圣的

483
00:18:20,400 --> 00:18:23,766
而是因为痛苦是最强的预测误差信号

484
00:18:23,766 --> 00:18:26,400
它在用最高的精度权重告诉你

485
00:18:26,433 --> 00:18:27,833
你现在的模型

486
00:18:27,866 --> 00:18:28,533
错了

487
00:18:28,833 --> 00:18:30,166
你愿不愿意听

488
00:18:31,400 --> 00:18:34,066
这期视频的内容如果引发了你的什么想法

489
00:18:34,100 --> 00:18:36,766
或者你有过自己的顿悟时刻想要分享

490
00:18:36,766 --> 00:18:38,400
欢迎在评论区留言

491
00:18:38,766 --> 00:18:39,666
你的故事

492
00:18:39,666 --> 00:18:43,433
可能恰好就是另一个人认知沙堆上的那最后一粒沙

493
00:18:43,900 --> 00:18:47,066
也欢迎告诉我你对下一期最期待的话题

494
00:18:47,733 --> 00:18:49,000
我是王利杰

495
00:18:49,133 --> 00:18:49,933
我们下期见