Try the demo and see for yourself!

A game that I hope will be fun. A game where you learn to see things faster and more dynamically.

The difficulty tuning is very much a work in progress.

This is a game where you advance by getting better at something. No collecting junk.

Inspired by the work of Sam Renshaw before, during, and after WWII. See also this paper and this copy of an article from the Saturday Evening Post editions 17th & 24th of April and 1st of May 1948 (I can’t testify to the accuracy of the transcript, but it’s a great article). In particular look at the first image before playing the game, and then after playing a few times and you’ll immediately see the difference

I’ve adapted the ideas to modern screens with much lower reliablity than Renshaw’s tachistoscopes (he trained some students to recognise a 7 digit number in only 3μs and claimed that almost anyone could be trained to recognise one in 0.5ms), but I’m hoping there will still be some advantage to memory, perception, and visual comprehension to be had with the 1/15 to 1/100 of a second duration that can be achieved on modern screens.

The game is playable, but in flux, so until this notice goes away, assume you’re seeing nightly material. Because you are.

Intended for use on a device with a keyboard. It might work on mobile, but don’t count on it. Mobile support is a goal, but not top priority.

The reason I started to write this was to see if I could train myself to widen my perceptual field. And to test Renshaw’s claims that this sort of training improves reading comprehension and thus indirectly improves reading speed.

With that in mind, let’s think about the structure of levels.

While playing this, my overwhelming impression is of being thrashed around. It’s kind of fun, but lots of variables are beings changed under the hood and I have limited control or insight into control them. That’s pretty bad considering I wrote the code last week.

The individual ideas: longer numbers, decreasing exposure times, multiple images, randomised locations, etc. are still good ideas, I believe, but I need to isolate them from one another at first and bring them together later on.

What’s the shortest exposure time for which I can reliably memorise an N digit number? That’s a useful metric.

Keeping score boards: you should be able to see your own stats. That requires keeping them. Local storage is good enough to start.

It’s a game after all, so how do we make it harder as you go?

This is currently implemented, and for the life of me I can’t get the 7 digit numbers in less than 1/15 of a second.

If you can’t memorise a given number size, we should show you numbers of that size for a longer duration and then work to shorten that duration.

If you don’t know where the image will pop up it requires a more gestalt approach to seeing, which is the point. Will help improve size of perceptual field.

Will this make a difference on a small screen?

Do we move the numbers each time you get it right, or every retry as well? Is that something that should scale with difficulty?

Put multiple numbers on an ellipse, evenly spaced but with a random initial angle. As the player improves, increase the size of the ellipse. The exposure time is too short to saccade accross the whole image, so you need to learn to see with the edges of the macula.

I should state now that I don’t know if such learning is possible, but it’s worth finding out.

Randomised delay prevents you from priming yourself. I think that’s important, but my only evidence is subjective.

Show columns of numbers to memorise. Or to really be a jerk just put N numbers at random points on the screen.

Add chaff to the mix. Put words or nonsense letters on the screen but you only need to memorise the numbers. This might just be frustrating.

Words, short sentences, even whole paragraphs with longer time lapse. The problem longer text with this is that we need comprehension tests since pure regurgitation isn’t what anyone wants.

What about nonsense text (jaberwock)? Garbage text (qweoasdafht)? You could probably publish a cool paper about memory profiles of real language vs syntactically/phonetically correct nonsense vs noise text. I don’t know if it would be any fun for players though.

Show 4+ images and make the player figure out which are the same. They can creep together like Pavlov’s circle & ellipse if I feel sadistic.

What bird did you just see? How many? which way were they going?

Basically replicated Renshaw’s work for the navy during WWII, but with birds instead of fighter planes.

I think this would be fun and useful, but I’ll need to find and label the images.

Assumes you care about birds. Could also use trees, mushrooms, fish, you name it. People who hunt mushrooms regulary see more mushrooms at a glance than you do in an hour of staring. There’s huge room for improvement for most of us, but maybe you’d do better turning this off and going out in the woods.