• View
• Edit
• History
• Print

ESLReactionGame

Day 1

Given the challenge to develop a kind of "proof-of-concept" use for the PicoBoard in conjunction with the iSENSE system, I decided that the PicoBoard button would be an appropriately straightforward input, and a reaction time game an appropriately straightforward application for that input.

I started off looking for a consistent stimulus for the reaction. Most reaction time applications on the web have random stimulus, so I looked for something like Guitar Hero, where the stimulus was consistent across trials but not regular within each trial. Ivan then suggested I just make my own stimulus in Scratch (of course!), so I wrote a Scratch program that flashed a screen between black and white. Once that was up, Fred suggested wiring up a photocell to the PicoBoard, then holding that up to the flashing screen to give iSENSE a measure for the stimulus, as well as a measure for the reaction.

After that, the network died.

Lessons Learned

• The PicoBoard cannot talk to iSENSE and Scratch at the same time.
• Coming up with interesting (i.e. individually and broadly meaningful) iSENSE exemplar experiments as proof-of-concept is a fun exercise but not that satisfying. Better to find people who have existing challenges, and try to figure out how to make iSENSE work for them.
• It is useful to have a backup offline version (even if just skeleton) for systems that are totally web-dependent, so that work doesn't have to stop if/when the network dies.

Day 2

The network was back. I got an experiment running and ran several trials with a very regular flashing stimulus. The differing scale was confusing, as was the noisiness of the photocell data. Francesca from OLPC then pointed out to me that I had designated the PicoBoard light sensor, rather than the photocell hooked up on port A, as the input for the flashing stimulus. Oops! I ran a new trial with the correct inputs and was much happier with the results.

Francesca wanted the ability to measure the lag time between the stimulus and the reaction and started to play with Scratch.

Lessons Learned

• Scaling of iSENSE visualizations is needed (CSV input helps, but it would be nice to have this within the iSENSE system itself)
• Designating the correct sensor inputs is an easy place for human error.
• Given a dataset, it can be easy to construct a plausible story even based on wrong assumptions. More eyes are better.