The Greatest Cognitive Science Experiment, Ever?

Chad Orzel has challenged the ScienceBloggers to come up with the greatest experiments in their respective fields. While Greta and I are reluctant to say this is the greatest experiment ever (there are so many great experiments!), we both independently came up with the same one: Roger Shepard and Jacqueline Metzler's 1971 experiment on mental rotation. It's certainly our favorite, and it's difficult to overstate its importance.

The design of the experiment is simple and brilliant; yet it was not easy to execute at the time. Today researchers studying vision almost always use computers to display stimuli. In 1970, when the experiment was designed, psychologists didn't have easy access to computers capable of displaying Shepard and Metzler's complex, three-dimensional images. They wanted to show participants two objects in two different orientations, then measure how long it took for them to determine if the objects were the same. Here are a few of the 800 different pairs of images they created for the study:


The objects were created using a computer program, but the computer's output was a set of coordinates, which then had to be plotted out by hand on graph paper. These images were transferred to cards, which were displayed to viewers using a tachistoscope, a viewing box with a shutter that allowed a precise measurement of when an image was revealed. Participants pulled one handle to indicate the objects were the same, and a different handle when they were different. Though half the objects were the same and half were different, Shepard and Metzler were only interested in the results when objects were the same.

The objects were either rotated in the picture plane so that the two-dimensional image did not change, or around the vertical axis (imagine the object spinning around a central pole), in which case the two-dimensional image was completely different. In both cases, the results were similar:


Though there are many possible explanations for the cognitive process that led to this result, the simplest is the one provided by the viewers themselves: to determine whether the objects were the same, they rotated one of the objects in their head and determined whether it matched the other object. This mental rotation occurred at a constant rate, so reaction times were longer when the rotation required to make the comparison was larger. The data matches this explanation perfectly. If the rate of rotation is constant, then after subtracting out the time required to make the comparison and operate the machine, then participants should take twice as long to make a 40-degree rotation as a 20-degree rotation, and three times as long for a 60-degree rotation. When you plot the results on a graph, the data points should line up in a diagonal line with a constant slope. And they do, all the way to the longest, 180-degree rotation. It didn't matter whether the objects were rotated in the picture plane or the seemingly more difficult vertical axis -- the results followed the same linear function.

So why is this experiment so important? In the early 1970s, the Sapir-Whorf hypothesis was taken quite literally -- the idea that we can't think about anything without language, that language itself shapes our thought. The Shepard and Metzler experiment, along with hundreds of similar studies that followed, demonstrated quite clearly that there are many aspects of cognition that don't involve language. Just try comparing the any of the two objects above using only a linguistic explanation -- your reaction times most certainly will not be related to how far the objects are rotated: instead, you'll have to say things like "the first object has three cubes in a row, followed by a right angle, then two more cubes..." then offer a similar description of the second object, and finally, compare these descriptions to see if they match. By this time, your mentally rotating friends will have completed dozens of trials of the same task.

In addition, the mental rotation paradigm has been used to study a number of other phenomena -- we've discussed how it relates to focal dystonia, for example. It has become one of the most powerful tools for studying the vision system, inspiring thousands of scholars.

Do you have a different opinion about the most important cognitive science experiment? Please let us know in the comments.

Shepard, R.N., & Metzler, J. (1971). Mental rotation of three-dimensional objects. Science 171, 701-703.

More like this

I've always liked this experiment, not only for the reasons you give, but also because the data is so clear cut. Didn't Shepard and Metzler calculate the "speed" of mental rotation per degree from this? How many other experiments culminate with such precise, borderline empirical numeric data?

By Jonathan Dobres (not verified) on 18 Jan 2006 #permalink

OK, it's definitely not the "greatest" experiment ever, but for the "brilliantly obnoxious" category I'd like to nominate Simons and Levin's Failure to detect changes to people during a real-world interaction. In a, well, brilliantly obnoxious study on change blindess, they sent an experimenter to walk around the Cornell campus and ask people for directions. In the middle of the conversation, the experimenter was surreptitiously replaced by someone else. Only half of the subjects realized they were talking to a different person. Best part: the surreptitious replacement occured when a couple of "construction workers" rudely carried a door between the experimenter and the subject. It's a little on the psych side of cog sci, but totally fun, no less.

By ThePolynomial (not verified) on 18 Jan 2006 #permalink


I have to agree; that's one of the most hilarious experiments of all time. We have reported on work by Simons and Levin (though not that experiment) here and here.

Was the Sapir-Whorf hypothesis taken literally by many people in the 1970s? Histories of the cognitive sciences that I'm familiar with tend to indicate otherwise. Strongly.

I also wonder if there are outliers (whether through innate ability or practice) whose ability to perform rotations like this is not correlated with degree of rotation.

I've called the Simons and Levin experiment the "coolest" experiment ever, but I think the best has to go to the original Stroop experiments, not only for the findings, and conceptual importance, but for the methodology. Stroop effects are everywhere, in every area of cognitive and social psychology.


I think you have a point there, but given that experiments were still being published on Sapir-Whorf effects in the mid 1980s, I'd say it was far from finished in 1971. (Though admittedly this would also suggest that Shepard and Metzler was not the final nail in Sapir-Whorf's coffin.) I'd appreciate it if you could give a couple references for the histories you mention.

Not to dodge the question, exactly, (though Gardener 1985), but is the Sapir-Whorf hypothesis consistent with any cognitive science per se?

Jonathan, there's been a rash of research over the last decade on linguistic relativity, and there's a growing consensus about the ways in which language affects both conception and perception. The short answer is, the research shows language does affect conception and perception.

Hello all on this comment board,
I'm currently looking into similar cognitive psychology studies as this one by Shepard and metzler, any suggestions? any that are easy to reproduce that i could perform on my own?

Mental rotation studies are difficult to reproduce on your own because they measure reaction time, which generally requires fairly sophisticated programming. I could imagine a replication where you show people pictures of two objects for a specified period of time, then ask if they are the same/different, and see if the accuracy of their responses are related to the degree of rotation.

And yes, there are literally thousands of studies on mental rotation. One of my favorites (for some reason) is this one:

Munger, M. P., Solberg, J. L. & Horrocks, K. K., (1999). The relationship between mental rotation and representational momentum. Journal of Experimental Psychology: Learning, Memory & Cognition, 25, 1557-1568.

You can also measure mental rotation using pen and paper if you focus on accuracy for lots of problems in a short amount of time. Check out

Vandenberg, S. G. & Kuse, A. R. (1978). Mental rotations, a group test of three-dimensional spatial visualization. Perceptual & Motor Skills, 47, 599-604.

There's an interesting laterality issue in mental rotation that isn't often noted. If you ask people which object (left or right) they rotate to make their judgment, right-handers report that they rotate the object in the right visual field and left-handers the one in the left field.
It's a strong effect and, unlike virtually all handedness tests, no one has ever been told how to mentally rotate.
(Cook, Frueh, Mehr, Regard & Landis, Brain and Cognition 25, 240, 1994).

By Norman D. Cook (not verified) on 20 May 2006 #permalink

Shepard & Metzler did a beautiful, elegant experiment, and their work is made all the more memorable because it is associated with those striking graphics, but it was certainly not what put paid to the assumption that thought is entirely (or mostly) dependent on language. Amongst psychologists, the realization that mental imagery plays a large and crucial role in thought and memory had been steadily growing for more than a decade at this time, and several scholarly books on the psychology of imagery, both monographs and collections of papers, had already appeared in the late 1960s and very early 1970s, before Shepard & Metzler's work had its impact. The most important figure in this movement (though certainly not the only one) was probably Canadian psychologist Alan Paivio, who, on the basis of an experimental program going back to the early 1960s, was able to build up a massive case for the importance of imagery as a form of mental representation. His major theoretical synthesis of this work, Imagery and Verbal Processes, appeared in 1971, the same year as the Shepard & Metzler paper (his latest work Mind and Its Evolution (2007) continues the same theme). Because of its elegance, its clear cut results (a nice linear graph), and its graphics, the Shepard & Metzler experiment caught people's imagination, but inasmuch as it contributed to persuading psychologists of the importance of non-linguistic forms of mental representation in human thought, it was, in fact, merely surfing the crest of an already very large and growing wave of opinion.

For more detail on this history (with references) see:

By the same token, Sapir's and Whorf's anthropologically inspired linguistic relativism was only one minor aspect of what Ian Hacking (1975) has called the "lingualism" (the idea that all thought, or all thought that matters, depends on language) that dominated the way intellectuals (not only psychologists, but philosophers and others) thought about the mind during most of the first half of the twentieth century. During that period, this attitude was widely seen as the ultimate in scientific hard-headedness. As far as psychology goes, the most important instigator of lingualism was surely not Sapir or Whorf (or the philosophical lingualists such as Frege Wittgenstein, and Austin), but J.B. Watson (who equated thought with sub-vocal speech). Certainly psychologists, like Paivio and Shepard, and others, who were involved in the revival of the psychology of imagery in the 1960s and '70s, quite explicitly presented themselves as reacting against the Behaviorist movement that Watson had initiated some 50 years before.

On this history see:

So yes, the Shepard & Metzler experiment is very beautiful, but its historical significance has been much exaggerated.

Hacking, I. (1975). Why Does Language Matter to Philosophy? Cambridge University Press.

Paivio, A. (1971). Imagery and Verbal Processes. New York: Holt, Rinehart and Winston. (Republished in 1979 - Hillsdale, NJ: Erlbaum.)

Paivio, A. (2007). Mind and Its Evolution: A Dual Coding Theoretical Approach. Mahwah, NJ: Erlbaum.