A Picture Is Worth A Thousand Words, Except When It Isn't

Sorry. Long time, no blog. I had a phone conversation a few days ago with David Kraemer, a post-doctoral fellow at the University of Pennsylvania's Center for Cognitive Neuroscience. To oversimplify, Kraemer is investigating the idea of "learning styles" (you know, the popular idea in education circles that we are either visual or verbal learners and that instructional materials should be so-catered ).  There are other "learning styles" out there, too, but visual and verbal are two mainstays. But, it turns out there's not a great deal of scientific support for this, says Kraemer, at least not for the theory as it's popularly perceived (i.e. learning styles appear someplace on a visual-verbal continuum, and so the more visual you are the less verbal, etc).

Anyway, Kraemer's adding spatial learning to the visual mix, as differentiated from "object" visual learning (a 3-D vs. 2-D distinction). And, lucky for me, he's testing all of these learning styles as they affect (or don't) our ability to navigate. When I was in Philly, I explored a bunch of desktop virtual cities for him after I filled out a few questionnaires meant to determine if I was more of a verbal or visual or spatial learner, etc. The city explorations were all videos of routes. I floating through a dozen or so deserted intersections, sometimes turning and sometimes not. The testing broke down like this:

1. View pictures of solitary buildings and say whether or not they were part of the city you just explored.
2. View pictures of intersections and say whether you turned right or left, or went straight through, or if the intersection wasn't part of the city you just explored.
3. View a picture of an intersection and then a picture of another intersection elsewhere in the city. "Point" using designated keyboard buttons to indicate what direction the second intersection is from the first.

Before I go into how I did, two caveats. Kraemer's research is still in the pilot phase. He's only tested a couple dozen people, and he's like to test maybe three times that many for good comparative validity. Also, the tests themselves have been evolving (up to and beyond when I did them). For example, my "learning phase" included three looks at each city, whereas some folks only went through each route once, others twice. I was the first to get the benefit of a third time through. On the other hand, in the building ID tests, the view-twice group was shown pictures of buildings as they actually appeared during the city exploration (identical orientation, etc, rather than the whole building in isolation that I had to ID).

Here's something a little spooky. I was the only subject who tested as an equally visual and verbal learner, and I thought for sure the verbal would be advantaged. I'm also really close on the object-spatial distinction, with a slight (surprising) advantage to spatial.

There are some very interesting, albeit tentative findings regarding visual-verbal-spatial learners and how they performed on each of the three tasks. But Kraemer's asked me to hold off on publishing anything until he's run a lot more subjects through the same tasks. So, I'll just report my own performance.

1. In landmark ID accuracy, I was better than all but two subjects in the "view once" group, which Kraemer said was probably a more valid comparison due to the advantage the view-twice group had in that task. But, full disclosure: I was tied for worst performance among the view-twicers.


2. In the intersection test, I pretty much mopped the floor with the view-oncers (I was about 92 percent accurate, and second best was about 52 percent), and I beat all but two of the view-twicers.


3. In the pointing task, what researchers call the "judgement of relative direction," and what's considered the most spatial of all the tests, I was in the top three compared to all the view-oncers and in the top four compared to the view-twicers.

Sure, I was probably a more motivated subject than the average undergraduate in it for $20, and I'll await  the results from more subjects to see if the trend of my above average performance continues. But I really think there's something to ponder here--either these tasks aren't really testing the abilities and skills required for real-world wayfinding, or this is evidence to support my suspicion that, despite how frequently I get lost, I have a very good sense of direction somewhere inside me that just needs to be properly tapped.

Interesting sidenote: Among the view-twicers, there was one subject that kicked everyone's ass, scoring tops in all three tasks and pointing like GPS, getting 65 percent correct in the judgement of relative direction task where the average (among everybody else) was about 22 percent. I hit 27 percent, btw. And that's right, I said HER score. This doesn't by itself mean the gender differences in navigation are all bogus, but it does fly in the face of stereotype.

A Garden of Forking Paths

I'm compelled to mention that I spent my first five minutes in Providence yesterday walking circles around the train station trying to get my world to fit the Google Map print-out I had for directions to the VENLab. Once I found the place, I spent a few minutes with Bill Warren, who heads up this virtual reality navigation lab to go over the day's schedule. Then to the VENLab itself, which is a 50 by 50 foot converted chemistry lab, empty, until you put on the VR helmet.

I began the day as a subject in an experiment by doctoral student Liz Chrastil, who is testing the effects on navigation of active vs. passive learning. She has people learn to find objects in a virtual maze--some by walking, some in a wheelchair with no control, some in a wheelchair with some control (i.e. telling the researchers where to push them), and some simply watching a recorded video of somebody else exploring the maze. Then, she has them go from one object to another, either through the maze or directly with the maze shut off and replaced with a grey expanse of desert.

With the helmet secured, and a transmitter cinched around my waist, I sat in a wheelchair (my random draw was to be a passive learner) and looked out into a leafy green hedge maze. Below was a gravel path and above was an unblemished blue sky. Crickets chirped through the headphones I wore, interrupted frequently by Liz's recorded instructions. Unlike most hedge mazes, this one had famous works of art hung on its walls by Dali, Monet, Magritte, and Renoir. I glanced at them as I was whisked through the maze and brought before the various objects hidden in its branches--a pedestal sink, a giant stone gear, an old well, a floating Earth, a book case, a clock, and a statue of a rabbit frozen in mid-leap (like a bad day in Narnia).

I didn't get a chance to ask the researchers how they managed to push me through a maze that they can't see, but "managed" is an apt word in this case, because I was frequently thrust through walls by accident and given glimpses of the desert world beyond the maze. Also, there's a slight delay in virtual reality between the movement of one's head and the rendering of the images, which messes with your vestibular sense enough to make many people a little sea sick. Being pushed through such a world only adds to the fun. Clip a few walls while you're at it, and you quickly become much more engaged with your stomach than your place in Euclidian space.

I sound like I'm making excuses, and maybe I am. Truth is, I have no idea how much my simmering nausea in the learning phase affected my performance in the testing. Probably, enough people have the same trouble that it should all come out in the comparative wash. Anyway, I was given the short-cut task, repeatedly wheeled through the desert to a spot where the wishing well or the bookcase or the sink,  would appear momentarily when I stood up and clicked a mouse, and then disappear again. Liz's recorded voice would tell me a second object. My job was to turn through the desert in the direction I believed the second object to be and then walk there (well, shuffle there. balance is tough in the virtual world).

I'll admit, this task was very difficult. And sometimes a little terrifying: The VENLab is equipped with a system to warn subjects enjoying a shuffle through a sunny virtual world that they're about to walk into a very real brick wall. It goes like this: An image of a brick wall springs into view and a stern, robotic voice yells out, "STOP! STOP! STOP! STOP! STOP!"

Good idea, I've rambled on far too long here. I'll catch up on the rest of my VENLab visit in the next post.

Pushing on the pull door

Since I began my lab-ratting, I've been filling out a lot of "rate your own sense of direction" or "rate your own spatial abilities" questionnaires. There are usually a series of statements, such as, "I very easily get lost in a new city" or "When I'm in a complex building, I can easily visualize what's outside the building in the direction I'm looking" and then you are asked to answer with varying levels of agreement ranging from strong agreement to strong disagreement.

Here's one of my favorites: "I always know if a chair will fit through my front door before buying it." Now, I've never bought a chair so large that there was any doubt that it would fit through my front door. But I do have experience trying to move a massive armoire into a second-floor apartment I was sharing with my sister. We had three people pushing and lifting, but the thing got stuck at a turn in the stairs. We managed to inch it up and forward through a series of twisting, lurching and cursing maneuvers, but after half an hour or so, we knew we were beaten. Then the real trouble started, because by this point we couldn't get it down the stairs either. It was pinned. We phoned an extra large friend for emergency help. He arrived just in time before my sister went to find an axe.

Here's another one: "It's not important to me to know where I am." It must be nice. But then again, aren't you curious?

And one more: "I enjoy reading maps." I do. I love maps. I like reading maps so much that I look at maps of places where I have no prospect of ever going. Likewise, I love historical maps (in the case of Boston, I like tracing the streets that once were bays, busy with merchant ships; I like to imagine the farm that was cut up in the late 1800s to create my neighborhood). Yet, I still get lost all the time. And my suspicion is that the assumption behind this question is that the more you enjoy maps, the more spatially oriented you are generally. So, it makes me wonder...

On and Off the Grid

I'll be making a return visit to the VEN (Virtual Reality Environment) Lab at Brown University tomorrow. Bill Warren, department chair and professor of cognitive and linguistic sciences, heads up the lab. He was kind enough to let me don the VR headgear and explore the virtual worlds of his lab back in 2003 for a Boston Globe feature I wrote on navigation research (my initiation into this research). Lucky for me, I had a researcher "wrangler" following me around while I explored to keep my feet clear of the wires connecting the VR headgear to the computers, and keep my face clear of oncoming brick walls.

Since 1999, Warren has used these virtual worlds to investigate all aspects of how humans "path integrate" (i.e. how well we track our own distance and direction over time as we move, independent of landmarks). A really good path integrator should be able to follow a circuitous path and take a shortcut back "home" and should, after enough exploration, start developing an accurate sense of how far and in what direction any location in an environment is from any other--in short, a Euclidian cognitive map.

A common test of path integration skill is to have people walk two legs of a triangle in a virtual desert, then have them turn and walk the third leg back to the starting point without any visual aids. After that, the researchers often start messing around... adding landmarks to see if that changes anything, and then moving landmarks to see if you notice, etc.

Generally, Warren has found humans to be pretty bad at path integrating. Recently, he's taken aim at the theory that we're ever capable of building up a Euclidian cognitive map. He's had people learn locations and routes in a virtual hedge maze and then secretly introduced "wormholes" within the maze that magically transport people to those locations. A Euclidian cognitive map would cry foul at such geometric dislocation, says Warren, but "nobody has found anything amiss."

Tomorrow, Warren will show me the wormholes. I'll also continue trying to diagnose my own sense of direction with the help of post-doc Mintao Zhao who is looking into what happens when path-integration and landmarks compete and with doctoral student Liz Chrastil who will run me through a battery of basic skills tests and have me take part in her active-vs-passive learning experiments.