Running Experiments 2: Response Hardware

We have the tools, we have the talent!
– Winston Zeddemore

In the last post I talked about timing of experiments in general, and mentioned that timing of responses is critical for success in a reaction time-type experiment. This topic will discuss some of the options for hardware that’s available for collecting responses.

The simplest solution is just to use the human-interface hardware which you’ve already got i.e. get your participants to click a mouse button, or a key on the keyboard as their response. However, there are several reasons why this will generally be undesirable. Most mice and keyboards on modern computers connect via USB, and this introduces a slight lag. The computer ‘looks at’ or ‘polls’ its USB-connected peripherals at a standard rate of 125Hz (meaning 125 times per second, or every 8 milliseconds). This means that if you make a response, there may be a (variable) lag of anywhere between 0 and 8 ms between the response and the computer actually ‘seeing’ it. USB-keyboards have a similar problem. In addition, mice (and keyboards) have a lot of internal circuitry which can also introduce timing lags of variable durations. This paper (Plant et al., 2003) presents the results of some bench-testing of a set of mice. The best mouse they tested had a minimum button-down latency of just 0.4ms, whereas the worst one showed a minimum lag of 48ms! Running timing-critical experiments with the latter mouse (where the effect you might be chasing could be in the range of 20-30ms) would clearly be disastrous.

High performance Razer gaming mouse

High performance Razer gaming mouse

So, some mice and keyboards might be suitable for running these kinds of experiments. Unfortunately, it’s hard to tell which peripherals are accurate without actually running some bench tests on them. This is not particularly hard, but requires some specialised equipment (say, a timing-accurate pulse generator and an oscilloscope). Doing a bit of research on different brands and models of peripherals can help as well. In recent years these issues have received quite a lot of beneficial attention from hardcore computer gamers. People who play online games at a high level are generally quite concerned with issues like mouse-lag, as a difference of a few milliseconds can mean the difference between victory and defeat against other players. Companies like Razer make specialised gaming peripherals which (they claim) minimise these timing issues. A quick search of google for “mouse lag” throws up a whole lot of tips and tricks from gaming forums which might also be useful for minimising these issues. Companies like this also make specialised game-pad devices (like the pads used for an Xbox or playstation) for PCs which are often pretty good too.

So, if you absolutely have to use a keyboard or mouse to collect responses: a) do some research, b) spend a bit of cash on the best gaming hardware you can find/afford and c) preferably, do some bench-testing of the hardware.

What other options are there? If you’re at all handy with a soldering iron and a circuit diagram (or if you know someone who is) then by far the cheapest option (and some might say, also the best) is to build a response box yourself. All the components you need are readily available from an electronics shop like Maplin or Radio Spares and are actually pretty simple. I used home-built response boxes for years with minimal problems. Care needs to be taken in the selection of switches – you need buttons which are firm enough so you don’t get accidental clicks by someone just resting their finger on it, but sensitive enough that they’ll record a deliberate click reliably. Buttons that are sensitive, but give a nice ‘click’ when pushed are the best – the best thing to do is go to a shop and try some out yourself. The usual way of building these things is to mount the circuit in a small plastic box, and connect the buttons through the parallel port of the computer. Inputs to a parallel port are ‘always-on’ so this gets around the USB-polling rate problem mentioned above. Circuit diagrams can be readily found online or in papers. Here’s an example for a four-button response box:

4-button response box circuit diagram, reproduced from Creeger et al., (1990)

4-button response box circuit diagram, reproduced from Creeger et al., (1990)

All it requires is a few buttons, a few resistors, and a few capacitors. Simples.

The final option is to use a specialised bit of hardware. There are several companies which make high-performance experimental hardware designed precisely to minimise the kinds of timing lags discussed above. Often, the producers of experimental software will also make custom built response devices specifically designed to interface with their software. Empirisoft will even sell you a ‘standard’ keyboard which has been gutted of its electronics and filled back up with their highly-timing-accurate circuit boards. Other companies which do this kind of thing are Psychology Software Tools (optimised to work with their E-Prime experimental software) and Cambridge Research Systems.

Six-button response pad made by CRS

Six-button response pad made by CRS

The downside of these is that they are generally eye-wateringly expensive. Also, whether you trust the manufacturers claims about timing accuracy is up to you – personally I wouldn’t use any of these for anything critical without doing some benchmarking first. But then, I’m a bit obsessive about these things.

So, that’s the hardware side of how to collect responses in experiments in an accurate way. The software side is a whole different ball-game, obviously, and I’ll cover some of those issues in a future post.

Happy experimenting!

References
Creeger, C. P., Miller, K. F., & Paredes, D. R. (1990). Micromanaging time: Measuring and controlling timing errors in computer-controlled experiments. Behavior research methods, instruments & computers, 22(1), 34–79.
Plant, R. R., Hammond, N., & Whitehouse, T. (2003). How choice of mouse may affect response timing in psychological studies. Behavior research methods, instruments, & computers : a journal of the Psychonomic Society, Inc, 35(2), 276-84.

Advertisements

About Matt Wall

I do brains. BRAINZZZZ.

Posted on February 26, 2011, in Experimental techniques, Hardware and tagged , , , , , , , . Bookmark the permalink. 6 Comments.

  1. If you only got one measure of an RT, a 0-48 ms lag would be disasterous. But RTs are inherantly variable anyway, so you will be collecting multiple measures of RT and from that calculating an estimate of the mean. Are you sure that extra equipment-related variability of this order of magnitude significantly affects the variability of your estimates of the mean?

    • It’s a good point, and one that I’ve never systematically sat down and done the maths to work out, because, well… life’s too short. Fortunately some other people have though:
      http://onlinelibrary.wiley.com/doi/10.1111/j.2044-8317.1989.tb01111.x/abstract

      This paper shows that even a very poor-resolution clock (30ms) has a pretty negligible effect on mean and variance of measured RT. So, you may well be right, even a fairly large equipment-related variability might not matter too much.

      However, I would say that RT data are generally pretty noisy and the variability derives from multiple component processes, so it seems unwise to ignore an additional noise source that could be (relatively) easily mitigated. Why use poor response hardware when you can build your own accurate response box with $20-worth of parts from Radio Shack?

  1. Pingback: TTL-Technology – Getting Things Talking to Each Other « Computing for Psychologists

  2. Pingback: BrainDesire, Online Experiments, Neuro-business and a Plea for Openness « Computing for Psychologists

  3. Pingback: Behavioural/Experimental software for psychology… A teaser. « Computing for Psychologists

  4. Pingback: How to make your own parallel port response boxes « Computing for Psychologists

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: