COGSCIdotNL
Home Optical illusions Side-scrolling Pac-man
 
Active vision and side-scrolling Pac-man
Active vision and side-scrolling Pac-man

In the video below, you see Pac-man being chased by a ghost (again)1. In the first part of the video, the maze scrolls so that Pac-man stays in the center of the screen (side-scroller perspective). In the second part, the maze stays put and Pac-man moves around (static perspective). Both perspectives are used in video games. The side-scroller perspective is used in many older platform games, such as Super Mario Bros, and is in many respects similar to the first-person perspective used in most action games today. The static perspective is used mostly in simple games and very old games, such as Snake and the original Pac-man.

So both perspectives offer perfectly acceptable gameplay. But the difference in your ability to detect changes is dramatic!

With the static perspective, you can see quite clearly that the maze is changing all the time: Except for a few tiles around Pac-man, the whole maze is randomly re-generated with each frame. However, with the side-scroller perspective these changes are much less noticeable, to the point that you may have failed to notice them at all, at least for a while.

So why does the perspective make such a big difference?

You might think that with the side-scroller perspective, there is already quite a lot of change going on, because the maze slides across the screen. The changing structure might simply get lost in the chaos. In contrast, with the static perspective, the maze structure is the only thing that changes, and is therefore much more noticeable. This is essentially an explanation in terms of change blindness, analogous to the one that I offered for a previous Pac-man-inspired illusion.

However, when you think about it, this cannot be the full explanation. After all, with the static perspective you track Pac-man with your eyes. Therefore, your vision is centered on Pac-man, just like it is with the side-scroller perspective. In fact, you might say that the only real difference is that the side-scroller perspective centers Pac-man for you, whereas with the static perspective, you visually center Pac-man yourself, by making eye movements.


Active exploration is also an important principle in autonomous navigation, such as used by the Mars rover Curiosity.

And this, I think, could be the crucial difference: The difference between actively tracking Pac-man as he moves across the screen, and passively observing Pac-man as he stays in the center.

In a classic essay, Donald MacKay suggested that making an eye movement2 is like asking a question. This sounds esoteric, but what he meant is the following: When you move your eyes, the way that the light falls onto your retina (the light-sensitive layer in the back of the eye) changes in a predictable way. For example, if you look directly at an object and make an eye movement away from it, you know that the object will still be there, but no longer be in central vision. This is a consequence of your eye movement that you can predict beforehand, and verify afterwards. MacKay’s ‘question’ refers to this principle: After every eye movement, your brain ‘asks’ whether the world looks as expected. If not, alarm bells go off, and a change is detected.

In the case of the Pac-man video, when you actively track the moving Pac-man, your brain is continuously making predictions and verifying that everything looks as expected. When an expectation is violated, in this case because the maze structure changes, you immediately notice that something is amiss.

In contrast, with the side-scroller perspective, your brain doesn’t make any predictions, because you do not make any eye movements. Your vision remains centered on the Pac-man, but not as the result of something you do. You are not in control. This lack of control means that you cannot predict what should happen. And the changing maze does therefore not violate any predictions: There are no predictions to violate.

In sum, I think that this video illustrates that seeing is an activity. We don’t just sit back and absorb information (and when we do, perception suffers), but actively track and scan the environment with our eyes. Seeing is like a continuous feedback loop of making and testing predictions. Or, to use MacKay’s terminology, of asking and answering questions.

References

MacKay, D. M. (1972). Visual stability. Investigative Ophthalmology, 11, 518-524.

  1. The song that you here in the video is Robot Wars by Binärpilot, which you can freely download from here.

  2. The argument holds for movement in general, but MacKay was primarily concerned with eye movements. As are we, in this case.

You may also enjoy these posts

Feel free to leave a comment! For lengthy discussions and technical questions/ remarks, please refer to the forum.