Many psychophysical experiments make use of the subject’s reaction time to a specific event or events as a way of measuring his response and it would seem that a computer system is an ideal way do this. For example, the keyboard could be used as a convenient way for the subject to indicate that he had seem something appear on the screen for instance. While good in theory this is more difficult than it seems. Most modern computers run multi-tasking operating systems, such as Windows95, which means that a user’s program may not in fact have control of the processor all of the time; if another task such as an autobackup wishes to access the hard disc it may steal quite large amounts of time from the other users which can mean that a key press will go unrecognised by the user’s program until control is subsequently returned and thus render the reaction time invalid. A further source of error lies in the keyboard itself which operates by scanning the rows and columns on a sequential basis. Typing is intrinsically a slow business so the scan speed may also be slow. This means that there is a variable time between hitting the key and the computer recognising it that depends just where it was in its cycle when the event happened. Unfortunately, similar arguments apply to other convenient input devices such as mouse buttons. The advice here really has to be that if you need accurate reaction times and are not sure exactly how the computer works you should devise some other external hardware to perform the measurement instead.

^ Back to top