The secret language of behavior

动物行为学的先驱廷伯根和洛伦兹依靠他们自己的眼睛和自己的大脑所做的观察,当他们提出动物行为可以分解成立体模式时。现在该字段已进入计算机时代,with the development of machine vision and learning technologies to extract scientific understanding from large video image datasets of untrained untethered animals.

The secret language of behavior

去年11月在圣地亚哥演讲Society for Neuroscience meeting,Sandeep Robert(Bob)Datta描述了他对小鼠行为的计算分析,目前仅发表研究在哺乳动物身上,reveals a modular structure.He emphasized the unsupervised nature of the machine learning he trained on the 3-D pose dynamics of  freely moving animals.   A subsecond structured pattern fell out of the data themselves – initially derived from videos of mice placed individually in buckets – but the same temporal patterning was detected in other environments as well.

对数据进行数学建模,确定了大约60个音节的曲目表,金宝搏体育each with a characteristic trajectory in pose space lasting on average 300 milliseconds,并由动态转换分开。改变环境-例如在竞技场的某个地点放置狐狸气味引起厌恶行为由相同的模块组成,但具有不同的频率和转换概率。

The study,在来自达塔实验室的视频(以下)中描述,现已在未发表的著作中扩展到演示如何通过这种方法解剖行为提供对药物和疾病突变的效应的敏感解读。

在一项实验中,500只小鼠中的每只被给予15种常用的精神药物中的一种,Datta's analysis using his model finds that each leaves a different behavioral fingerprint,对一只小鼠的特定药物进行诊断。

最近,他对两个自闭症基因小鼠模型(cntnap2/)和16p11.2的行为表现型进行了研究。在一系列传统的行为分析后报告为高活性。,reveals clear differences in the alterations of behavior shown by the two mutant strains.In each case there are differences with wild-type mice in the usage of 8 specific syllables,但是除了一个重叠,受影响的音节对于这两个遗传模型是不同的。

分析还提供了对利培酮对小鼠CNTNAP-2表型有疗效。是一种FDA批准的用于治疗易激性孤独症(精神分裂症和躁郁症的躁狂期)的药物。在受cntnap2缺乏症影响的8个音节中,只有一个同样受到利培酮的影响,while other normal syllables are also used less frequently.The results imply that the drug,至少在老鼠身上,is acting as a sedative rather than correcting circuit defects caused by the mutation.

With a company spin-of  (音节生命科学)利用这种新的能力来破译小鼠的肢体语言,以获得治疗效果,这种新的能力已经在进行中,而对于研究人员来说,达塔的模型是免费提供的(一个声明受到研讨会与会者自发鼓掌的欢迎)。

Linking behavior to neural circuits

有许多其他的演示涉及计算行为学会议,and several illustrated how automated tracking can be combined with modern tools for monitoring and manipulating neural circuits to give insights into neural circuit function.

研究果蝇果蝇果蝇,具有高度发达的遗传工具的模型器官,在这里开辟一条小路。

In earlierpublished work从他在普林斯顿Joshua Shaevitz实验室当博士后开始,与Bill Bialek合作,戈登伯曼用无监督的机器学习来探究果蝇的地面社会行为。identifying 117 stereotypic motifs,平均持续0.21秒和通过平均持续0.13秒的过渡连接。

这些数据的数学建模indicates a hierarchical organization over different time scales,与目前果蝇神经回路组织模型一致。伯曼现在正在解决一个难题,即在颈部(比大脑或身体其他部位的神经元数量少100倍)的向下轴突处,破译这些行为的神经指令代码。collaborating with Josh Shaevitz,杰西卡·坎德Gwyneth Card和David Stern_to_analyze_the behavioral effects of optogenetical importing these axons.

感谢Janelia Farm为研究界开发和共享工具的使命,there are 2,215 GAL4 Drosophila lines that can drive the cell-type specific expression of genetic constructs,such as the light- or heat-activatable channels used in opto- and thermogenetic experiments.

A neat example of applying these technologies was given by Brian Duistermars,who has tracked and analyzed the threat displays of male fruit flies,将对这种行为的控制固定在6个神经元上,这些神经元能够根据刺激它们的强度来指导全部或部分行为。

A broader,对2中热遗传激活后行为改变进行更为系统的筛选,200 GAL4 lines was described by Kristin Branson.她的团队发展了贾巴,an interactive machine learning approach that incorporates an element of supervision from biologists in analyzing animal behavior,并以此对果蝇20种不同的地面社会行为进行分类,其修改可以通过计算检测到。

这使得大脑行为图的生成将高维行为数据与整个神经系统中特定神经元的操作联系起来。虽然结果尚未公布,这个project实验室视频(以下)中的特点和珍妮亚的传统,the group is working on ways of sharing the data through a可浏览地图集行为-anatomy map.

斑马鱼是通过计算行为学产生新见解的模式有机体的另一个例子。

Gonzalo de Polaviejo studies decision making by zebrafish in a social context,using his发布的ID跟踪软件它们相互作用时能可靠地识别个别鱼类。The fish engage in non-random social interactions from 5 days post-fertilization and develop their social behaviour while still transparent larvae,有希望建立一个有用的模型,在该模型中寻找神经活动的相关性,并测试基因突变和药物的影响。

But mice remain the dominant model for biomedical research relevant to human disease.

Gordon Berman uses his unsupervised machine learning and modeling approach to study behaviour in mice and humans as well as Drosophila,并且使他的可用型号on Github.Preliminary results on the behavior of mice in both social and nonsocial contexts,本文介绍了与刘瑞敏团队合作取得的成果。poster.

梅根·凯里机车to analyze the limb,head,自由行走小鼠的尾巴运动学,并研究干扰小脑电路的影响。

Azim Eiman uses unsupervised learning from machine vision to detect and reliably quantify subtle variations in reach and grasp behavior as he builds on hispublished work为了理解小脑电路是如何与运动指令相交的,从而微调这些面向目标的运动。

Adam Hantman applies the interactive machine-learning approach developed by Branson as he analyses the role of the motor cortex in controlling reach and grasp behaviour in head-fixed mice.His出版作品令人印象深刻”停顿和玩耍利用光遗传学的操纵水平,and he is now looking to correlate the kinematics of the behavior with activity in the motor cortex and subcortical areas,最近,由Sofroniew et al 2016,实现不同脑区同时钙成像,单神经元分辨率。

Overall,the message is that ethology is on course to catch up and connect with advances in monitoring and manipulating neural circuit activity. With the aid of sophisticated computational analyses,我们开始理解行为的语言,大脑的最终输出。

188bet官网

评论