11/03/2012

3 interesting "embodied cognition" papers that aren't actually embodied


In previous entries, I've presented some pretty convincing evidence that, contrary to the popular idea of a mind-body split, body movements and sensory data can significantly affect the way we think. A series of studies by Spivey and Lleras suggest that eye movements in particular might be linked to cognition in an interesting, but indirect, way. In one of these papers, they write,
"When constructing or interrogating a spatial mental model, eye movements...are used to coordinate elements of the internal model with elements of the external world...the fact that these eye movements are not directly driven by any external visual input...[points to] an embodied mind that naturally activates 'lower-level' motor processes to accompany 'higher-level' cognitive processes because, rather than being separate functions that are triggered after the instantiation of a mental state, motor processes are intrinsic components of the mental state."

So what's special about eye movements? Do they really influence higher-level cognition, and if so, how do they do so? Let's review the evidence and find out.



Study 1 (Spivey & Geng, 2001):*
This study involves two very different experiments which both show a "surprising" link between eye movements and some sort of higher cognition. In the first experiment, participants heard a verbal description of a scene.  Their eye movements while imagining what they heard matched the ones they would have generated if they were actually viewing the scene. In this case, eye movements occured unprompted by any outside visual stimuli. In the second experiment, participants looked at a screen with several shapes on it that differed in color and direction of tilt. One of the shapes disappeared, and they were asked about either its color or its tilt. Although looking at the empty space where the shape once was provided no direct information, participants still looked there when asked about tilt. In this case, eye movements seem to act as a retrieval cue for remembering the missing shape.

In the first experiment, six participants' eye movements were recorded while listening to stories like the following:

Upward story: "Imagine that you are standing across the street from a 40 story apartment building. At the bottom, there is a doorman in blue. On the 10th floor, a woman is hanging her laundry out the window. On the 29th floor, two kids are sitting on the fire escape smoking cigarettes. On the very top floor, two people are screaming."

Leftward story: "Imagine a train extending outwards to the left. It is pointed to the right, and you are facing the side of the engine. It is not moving. Five cars down is a cargo holder with pink graffiti sprayed on its side. Another six cars down is a flat car. The train begins to move. Further down the train you see the caboose coming around a corner."

Control story: "Imagine you are on a hill looking at a city through a telescope. Pressing a single button zooms a specific block into view. Another button brings a gray apartment building into focus. Finally, a third button zooms in on a single window. Inside you see a family having breakfast together. A puppy appears and begs for a piece of French toast."

To prevent participants from guessing the hypothesis, they were told they were taking a break from the "real experiment" and that the eyetracker was turned off. The study found that participants tended to make more eye movements in the direction specified by the story (even though none of these studies explicitly used the words "up," "down," "left," or "right"). In other words, they behaved as if they were viewing the scene, not just imagining it. Spivey & Geng conclude that "a construction of a mental image is almost 'acted out' by the eye movements, and a mental search of internal memory is accompanied by an occulomotor search of external space." Although they hedge a bit later, they want to claim that the eye movements don't just automatically come with imagining a scene. Rather, they play a role in generating the scene itself.

This finding is not as revolutionary as it sounds at first. We already know that brain activity in visual areas while imagining a scene is almost the same as brain activity while viewing it in real life. The new wrinkle here is that eye movements are also the same. This should not be surprising: if the brain can't tell the difference between imagination and reality, then it will direct the same eye movements at an imagined scene as a real one.

In the second study, 32 participants looked at a computer screen with a shape in each corner. Each shape had a particular color and was tilted in a particular direction. Participants were told to look at each location and then look at the center of the screen. Then the display vanished and reappeared with one of the objects missing. Participants were asked either what color the missing object was or which direction it was tilted. An eyetracker was used to figure out whether participants looked at the blank area where the shape used to be. According to Spivey and Geng, if participants look at the area even when they got no information from it, then the eye movement must somehow be helping them access their representations of the missing shape. Thus, eye movements must be able to influence memory.  (Of course, if eye movements didn't influence memory but still automatically accompanied it, they'd have gotten the same result).

When participants weren't looking at the center of the screen as instructed, their eyes moved to the area where the object used to be (or about 24% of the trials). Interestingly, they were more likely to look at the object's former location following tilt questions than color questions. This may be because tilt is a spatial property of an object, and is processed in a similar area in the brain to object location (the "dorsal visual pathway" in the parietal lobe). Thus, the location of the object on screen might provide a cue for remembering how it was tilted. By contrast, color is processed in a different system (the "ventral visual pathway" in the temporal lobe), and thus would not be expected to benefit from a spatial cue. It's also possible that color is easier to notice and remember than tilt, so participants are less likely to feel the need to "look back" at the object when asked about it. In short, it seems that people unconsciously use eye movements as a memory cue, but only for some kinds of information.

The commonality between these two sub-studies is that eye movements can occur without being directed at anything in the external environment. Rather, they seem to move with a person's attention.

Study 2. (Grant & Spivey, 2003)*
If eye movements only correlated with boring problems like the ones in the previous study, one would be justified in not caring. This study, however, shows that eye movements can help people solve insight problems (problems that are about as "high level" as you can get). In other words, by using visual cues to cause the eyes to move in a particular way through a diagram, researchers can actually double the number of participants who solve an insight problem problem.

The authors used a classic and notoriously difficult problem, Dunker's radiation problem. In it, you're asked to use lasers to cure a person with an inoperable stomach tumor. When these lasers reach a certain intensity, they destroy all organic tissue. How can one use lasers to destroy the tumor without killing the tissue surrounding it? This problem is fiendishly difficult for most. When I first saw it (in a class several years ago), I wasn't able to solve it. (The answer is to fire several low-intensity lasers from different angles so that they converge at the tumor, with a combined intensity strong enough to eliminate it).

Subjects had 10 minutes to solve this problem while looking at a diagram, and they wore an eyetracker.

In the 30 seconds before coming up with the solution, participants who solved the problem on their own spent significantly more time looking at the skin area of the diagram than those who only solved it with hints. The groups did not differ in how much they looked at any other area of the diagram.

Once they identified this critical area, the authors wanted to know whether they could make people solve the problem more often by making them look at the skin. If they could, it would seem to imply that this sort of eye movement helps participants solve the insight problem.

Eighty-one undergraduates saw the tumor diagram, which either had an animated tumor that "pulsed," animated skin that "pulsed," or had no animation.   The researchers expected drawing attention to the skin with the animation would lead more students to solve the problem.  It worked.  About twice as many participants spontaneously solved the problem when the skin was animated (67% vs. 37%).  The animated tumor had no effect.

The authors looked at participants' eye movement problems while solving the problem and found that those who figured it out looked back and forth between the outside and the tumor, crossing the skin as they went.  In other words, their eyes followed the same path as the laser would need to follow.  Participants who needed hints showed the same pattern after they got their hint.  One explanation--the one I thought of--might be that the eye movements reflect an understanding of the problem that has not yet reached conscious awareness.  The authors suggest that the in and out movements of the eyes may have led participants to imagine multiple rays, which in turn gave them the solution to the problem.  There may be other explanations as well.  This study isn't set up to differentiate them, but the bottom line is that there is an intimate relationship between eye/attention movements and problem solving.

Study 3 (Thomas & Lleras, 2009):*
The authors wanted to know whether it was the eye movements that helped people solve insight problems, or simply their shift in attention.  So they had 92 participants solve Duncker's radiation problem with regular breaks in the middle to do a different task.  In one condition, while the problem diagram remained onscreen, they tracked a string of digits as it moved across the diagram in in-and-out patterns.  In another, while the problem diagram remained onscreen they attended the string, but kept their eyes fixated in the center of the display.  If they moved their eyes, they were reminded to maintain fixation.  In the final condition, while the problem diagram remained onscreen the entire string appeared at the center of the display near the tumor (they moved neither their attention nor their eyes).  Some participants in the final condition were asked to keep their eyes fixed center for the entire experiment.

In other words:
1) Move eyes and attention.
2) Move attention, don't move eyes.
3) Don't move eyes or attention.

People in conditions 1 and 2 were more likely to solve the radiation problem than people who moved neither their eyes nor their attention (Condition 3).  There was no significant difference between people who moved their eyes and attention and people who only moved their attention.

Bottom Line
Taken together, the findings of all three studies are consistent with the idea that the driving causal factor is spatial attention--attention that focuses like a laser on a particular point in space and can move through a scene. When participants' eyes moved as they imagined verbally presented scenes, these movements were driven by their attention moving through the scene. When participants looked at the blank area where an object of interest used to be, they did so because their attention was directed at that object and associated it with that space (at a level of consciousness below the one that knows looking at a blank space is uninformative). And of course, when participants' eyes move in and out of the diagram for Duncker's Radiation Problem, it's because their attention is jumping back and forth.

Now, you may ask: if attention is the crucial factor, why do people bother moving their eyes at all? If you've ever participated in an experiment where you've been told to move your attention without moving your eyes, you'll find that it's hard. Our eyes need to move with our attention so we can see sudden threats, like a predator sneaking up from the periphery.

These look like embodied cognition papers on the surface, but they aren't when you look at them closely.  There's nothing spooky or mysterious about eye movements. They just happen to come along with the real driving force for solving spatial problems: the attentional spotlight.

References:
*Spivey & Geng (2001). Oculomotor mechanisms activated by imagery and memory: eye movements to absent objects. Psychological Research 65: 235-241
*Grant & Spivey (2003). Eye movements and problem solving: Guiding attention guides thought. Psychological Science, 14:5, 462-465
*Thomas & Lleras (2009). Covert shifts of attention function as an implicit aid to insight. Cognition 111: 168-174