中文导读

我们学会分辨熟和不熟的水果、判断医学影像上的斑点是否危险时，通常以为是大脑在做判断，身体只是事后执行决定。几十年来，关于“分类学习”的研究也大多这样看待动作：人先做决定，再按下按钮。美国佛罗里达大西洋大学（Florida Atlantic University）10 月 24 日经 Phys.org 发布新闻稿，介绍该校 Luke Rosedahl 与布朗大学研究者发表在 Science Advances 上的研究。团队设计了虚拟现实实验：参与者学习判断图案属于哪一类，有的只按按钮，有的用大幅度手臂动作作答，比如挥动虚拟光剑击中目标，或向与图像分开的应答框出拳。结果很有选择性：对需要同时整合多种特征、凭整体模式判断的类别，大幅度动作明显提高了学习效果和正确率；而对按颜色、形状等明确规则分类的任务，即使加大难度也几乎没有帮助。无论是击打目标还是击打另一个框，效果都出现，说明起作用的是动作本身。补充实验控制了看图时间和反应速度后结论不变，而且用大动作训练过的人改用按钮后仍保持优势。研究者认为这或许能用于医学诊断、飞行等复杂技能训练，但研究只在虚拟现实中进行，现实场景中是否同样有效还需要更多研究。文章按“常见看法—实验—结果—边界”展开，阅读时可留意 however、whether、not just 如何表达转折、条件与比较，以及 suggests、may、seemed to 如何为结论划定边界。

正文

Is learning something that happens only in the head? When we learn to tell a ripe fruit from an unripe one, we usually picture the brain doing the work while the body simply acts on the decision afterwards. A new study suggests the body may play a bigger part than expected, but only in some kinds of learning.

本段译文

学习只发生在头脑里吗？当我们学习分辨熟水果和没熟的水果时，通常会想象是大脑在做这项工作，而身体只是在事后执行这个决定。一项新研究表明，身体发挥的作用可能比人们预想的更大，但只在某些类型的学习中如此。

The research was carried out by scientists at Florida Atlantic University and Brown University and published in the journal Science Advances. Florida Atlantic University described the findings in a news release on October 24. For decades, studies of how people learn to sort things into groups have treated movement as the last step: a person makes a decision and then presses a button.

本段译文

这项研究由佛罗里达大西洋大学和布朗大学的科学家完成，发表在《科学进展》期刊上。佛罗里达大西洋大学在 10 月 24 日的新闻稿中介绍了这些发现。几十年来，关于人们如何学习把事物分组的研究一直把动作当作最后一步：人先做出决定，然后按下按钮。

To test this idea, the team built a virtual reality experiment. Participants looked at patterns and learned which group each one belonged to. Some answered with simple button presses. Others used large arm movements, either swinging a virtual lightsaber at a target or punching a response box that was separate from the image.

本段译文

为了检验这一看法，研究团队设计了一个虚拟现实实验。参与者观看一些图案，并学习每个图案属于哪一组。有些人只用简单的按键作答。另一些人则用大幅度的手臂动作作答，要么挥动虚拟光剑击打目标，要么向一个与图像分开的应答框出拳。

The researchers tested two types of learning. In one, people could follow clear rules, such as sorting objects by colour and shape. In the other, they had to combine several features at once and recognise a pattern, often without being able to explain how they decided. Judging whether a mark on a medical scan is dangerous or harmless is an example of this second type.

本段译文

研究人员测试了两种学习。在一种学习中，人们可以遵循明确的规则，比如按颜色和形状给物体分类。在另一种学习中，他们必须同时综合几个特征并识别出一种模式，而且往往说不清自己是怎么做出判断的。判断医学扫描图上的一个斑点是危险还是无害，就是第二种学习的例子。

The results were surprisingly selective. People who made large movements learned the pattern-based groups better and gave more correct answers than those who pressed buttons. However, the movements brought no real benefit for rule-based learning, even when that task was made harder. The benefit appeared whether people hit the target or punched the separate box, so the movement itself seemed to matter.

本段译文

结果显示出令人意外的选择性。做大幅度动作的人比按按钮的人更好地学会了基于模式的分组，答对的也更多。然而，对于基于规则的学习，这些动作并没有带来真正的好处，即使把这项任务变得更难也是如此。无论人们是击打目标还是击打那个分开的框，这种好处都会出现，所以起作用的似乎是动作本身。

In further tests, the team controlled how long people saw each image and allowed for differences in how fast they answered, and the result stayed the same. People who trained with large movements also kept their advantage after switching to button presses. This suggests the movements helped learning itself, not just answering at that moment.

本段译文

在进一步的测试中，研究团队控制了人们观看每张图像的时间，并考虑到了人们作答速度的差异，结果依然不变。用大幅度动作训练过的人在改用按键后也保持了他们的优势。这表明这些动作帮助的是学习本身，而不仅仅是当时的作答。

Luke Rosedahl, an assistant professor at Florida Atlantic University who worked on the study, said movement could offer a simple way to improve training for difficult skills such as medical diagnosis and flying. Still, the study took place in virtual reality, and the researchers say more work is needed to find out whether the benefit appears in real-world settings.

本段译文

参与这项研究的佛罗里达大西洋大学助理教授 Luke Rosedahl 表示，动作或许能为医学诊断和飞行等困难技能的训练提供一种简单的改进方法。不过，这项研究是在虚拟现实中进行的，研究人员表示，还需要更多研究来弄清这种好处是否也会出现在现实场景中。

新闻来源

• Phys.org (Florida Atlantic University) · Swinging virtual lightsabers and punching improve pattern-based learning in VR tests (24 Oct 2026)
• Science Advances · Learning by moving: Physical motion accelerates category learning (DOI: 10.1126/sciadv.aed4003)

✳更多文章 →