Qazi Rahman is the author of a recent paper entitled "Sexual orientation-related differences in allocentric spatial memory tasks". Here's a short interview with Rahman about the paper and its findings.

What kinds of tests did you use in this study and what do they evaluate?

We used two common tests of spatial learning and memory, the Morris Water Maze (MWM) and the Radial Arm Maze (RAM), which were designed for animals and made into human analogues using virtual reality by colleagues at Yale University.

The MWM tests the ability to find a hidden platform submerged under the virtual pool. Subjects start from a different place each time, so it tests the ability to form cognitive maps independent of your own body perspective using external cues.

The RAM test measures people’s ability to find rewards in the arms of a maze. RAM is more generous in allowing error in the use of external cues to locate the goal. The number of possible ‘routes’ in the MWM is much larger since there are no confined passageways like in the RAM.

Here are a couple of small screenshots of the virtual environment they used for testing:

What did you find?

In the MWM we found that gay men and heterosexual and lesbian women took longer to find the hidden platform compared with heterosexual men. The differences are large, almost one standard deviation of a difference.

Heterosexual men were faster from the first trial and maintained this advantage across a block of trials. However, gay men and heterosexual men spent more time in the correct area of the virtual pool compared to the female groups, suggesting that gay men are female-like in their spatial learning and ability to update information depending on where they are, but they are similar to heterosexual men in their eventual spatial memory for that information.

In the RAM test we found that both gay and heterosexual males had a greater ability than females overall.

That's interesting, but I'm wondering about their subject pool. I've requested a copy of the paper, and I'll revisit this post after I have a closer look at it, but one thing that strikes me is that the virtual environment is very much like a video game. Couple that with the tendency for psychological experiments to usually draw from undergraduates (~20 yrs. old) and the fact that young males are more likely to play video games, I'm wondering if they controlled for this. It may just be a bias of familiarity with the testing conditions, because males play so many more video games. One way to tell would be to screen subjects based on their experience with video games. Another would be to rerun the experiments in real, rather than virtual, environments.

Anyway, I'll post another update to this research when I find out more.

UPDATE: I've got a copy of the paper now, and they didn't use undergraduate subjects. Instead, they recruited from the local population with advertisements. They had 140 participants ranging in age from 19-45. They used "snowball sampling", which means that initial recruits were asked to recruit others (presumably their friends). So this could lead to subject with common interests and backgrounds. So I still think different experiences with apparatus and environment could be an issue, and I'd like to see some more controls.